I am attempting to fit a partial credit model using the PCM() function in eRm. My data set comprises 88 items in 43 individuals with possible scores 0,1,2,3.
Attempting either a PCM or RSM model fit simply hangs - no error messages but no return to command prompt and a continuously spinning "work in progress" indicator. The scores are from an instrument whose relevant psychometrics (e.g unidimensionality) have been established in other clinical populations: I'm studying a new disease group. I strongly suspect the problem relates to undesirable properties of the dataset: a significant missing data rate (about 20%) and a rather uneven distribution amongst the possible scores with relative paucity of 1 and 2 responses. I have in fact already deleted one item because of an "The following items do not have responses on each category" warning. I was trying to generate some diagnostics to get a better idea of what the problem(s) are, and/or adjust default parameters used in the nlm call to see if this helped achieve convergence. Setting options(warn=1) doesn't appear to help and am unclear how set flags such as print.level for the nlm call? Thanks for any help Rob Forsyth ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
